I must ask a provocative question. Is JupyterLab a hobby project or do experts work seriously on this project?
JupyterLabs architecture seems inappropriate or not existing. I don’t know what level of experience the developers have, but their issue discussions about the use of webpack and other topics make them appear like newbies.
Many people have asked why the development cycle is so extremely slow. Nothing has improved except for trivial fixes.
Documentation for extension developer is basically the “Let’s make a … extension” tutorial. No architecture overview, not even 10% complete. No cookbook for common problems. Everybody has to guess out of nowhere, how to tame this beast. Only source of information is this forum. Well, if I wait 1 minute for every code change to appear I can also wait 2 days whether somebody replies to my forum question.
- Feature freeze
- Engage all “developers” into documentation writing
- Start documenting the phosphorjs design
- Document the jupyterlab design.
- Find an expert to replace your architecture with something efficient (hint: modularization)
- Update the documentation
- If all this is too big for you, stop it.